The role of ROC75 as a daytime component of the circadian oscillator in Chlamydomonas reinhardtii

The circadian clocks in chlorophyte algae have been studied in two model organisms, Chlamydomonas reinhardtii and Ostreococcus tauri. These studies revealed that the chlorophyte clocks include some genes that are homologous to those of the angiosperm circadian clock. However, the genetic network architectures of the chlorophyte clocks are largely unknown, especially in C. reinhardtii. In this study, using C. reinhardtii as a model, we characterized RHYTHM OF CHLOROPLAST (ROC) 75, a clock gene encoding a putative GARP DNA-binding transcription factor similar to the clock proteins LUX ARRHYTHMO (LUX, also called PHYTOCLOCK 1 [PCL1]) and BROTHER OF LUX ARRHYTHMO (BOA, also called NOX) of the angiosperm Arabidopsis thaliana. We observed that ROC75 is a day/subjective day-phase-expressed nuclear-localized protein that associates with some night-phased clock genes and represses their expression. This repression may be essential for the gating of reaccumulation of the other clock-related GARP protein, ROC15, after its light-dependent degradation. The restoration of ROC75 function in an arrhythmic roc75 mutant under constant darkness leads to the resumption of circadian oscillation from the subjective dawn, suggesting that the ROC75 restoration acts as a morning cue for the C. reinhardtii clock. Our study reveals a part of the genetic network of C. reinhardtii clock that could be considerably different from that of A. thaliana.

In addition, a considerable number of genes, which are not conserved in the A. thaliana genome or are conserved but have not been described as a clock component, are known to be involved in the circadian clock of O. tauri and C. reinhardtii. These genes include LOV-HK and CPF1 of O. tauri [27,28], as well as ROC55, ROC59, ROC114, NAT3, XRN1 (ROC86), C1, C3, and pCRY of C. reinhardtii [25,[29][30][31][32]. It is noteworthy that cryptochrome genes (CPF1 and pCRY) have a critical role in the circadian oscillators of these alga [27,32]. Cryptochromes are the core components of the mammalian circadian oscillator [3]. While mammalian cryptochromes and CPF1 belong to a phylogenetic clade called "animal-CRY," pCRY belongs to "plant-CRY" with the A. thaliana cryptochromes, CRY1 and CRY2, which are involved in the input of light information to the circadian oscillator but not in the oscillator itself [33,34]. Taken together, it is likely that the chlorophyte algae have developed their own mechanisms different from angiosperms to adapt to their environment during evolution.
Here, we investigated the network architecture of the circadian transcriptional loop of C. reinhardtii to shed some light on the evolution of circadian clocks in green plants. We focused on the GARP protein gene, ROC75, in C. reinhardtii. ROC75 shows daytime mRNA expression [25], whereas ROC15, the other clock-related GARP protein gene in C. reinhardtii, shows nighttime mRNA (and protein) expression [35] similar to LUX (PCL1)/BOA (NOX) in A.thaliana [10,11].

roc75 is a conditional mutant
Previously, we had isolated roc75, which shows arrhythmicity under constant darkness (DD) [25] in the bioluminescence rhythm derived from the tufA-lucCP reporter gene introduced into the chloroplast genome [36]. Here, we tested the bioluminescence reporter rhythm under continuous light (LL) conditions using various light intensities (50, 10, and 2 μmol/m 2 /s) and in a light/dark (LD; 12-h light:12-h dark) cycle. Under the LL conditions, roc75 showed bioluminescence rhythms with lower amplitudes than those of the wild-type (WT) strain at all the light intensities tested (Fig 1A and 1D), indicating that ROC75 is required for the normal rhythmicity of the reporter under LL conditions, as well. However, roc75 showed rhythmicity under the LL conditions unlike the arrhythmicity it showed under the DD conditions ( Fig 1B  and 1D). In addition, a diurnal bioluminescence rhythm, which appeared normal, was detected in roc75 under the LD condition (Fig 1C and 1D). This rhythm cannot be explained by a simple light/dark response of the reporter, since another arrhythmic mutant roc114 [25] showed a lower amplitude of the bioluminescence rhythm under the same condition (S1A and S1B Fig). Taken together, these results indicate that ROC75 is indispensable for circadian rhythmicity under DD conditions but not in the presence of light. This implies that the function of ROC75 is compensated by light to some extent.

ROC75 is a day/subjective day-phase-expressed nuclear-localized protein
We analyzed the temporal and spatial expression profiles of ROC75. As reported in a previous study [25], the expression of ROC75 mRNA was high during the early subjective day under LL conditions (Fig 2A). A similar circadian profile was observed under DD conditions (Fig 2A). Under the LD cycle, the ROC75 mRNA showed a sharp peak at dawn (Fig 2A). The acute decline after the peak was due to a downregulation by light (Fig 2B). These mRNA rhythms were blunted for the mutated ROC75 mRNA (having a large insertion of the aph7" gene [25]) in roc75 (Fig 2A).
To analyze the expression of the ROC75 protein, we introduced a codon-adapted hemagglutinin (HA) epitope tag sequence [35] at the end of the coding sequence of a genomic DNA fragment of ROC75 locus, and roc75 mutant was transformed with the DNA fragment (S2A Fig). The tagged gene fully complemented the arrhythmicity of roc75 under DD conditions, indicating that the HA-tagged ROC75 was functional (S2B Fig). The ROC75-HA protein showed a rhythm, peaking at the subjective day under LL conditions ( Fig 2C). Furthermore, bioluminescence assay by using a luciferase reporter translationally fused to the C-terminal of full length ROC75 protein (ROC75-LUC) [35] and the luciferin analog AkaLumine (see Materials and Methods) also indicated diurnal/circadian rhythms peaking at the day/subjective day under LL, DD, and LD conditions (Fig 2D). These results suggest that ROC75 is a day/subjective day-phase-expressed protein. In addition, because ROC75-LUC complemented the arrhythmicity of roc75, the ROC75-LUC fusion protein is probably functional as a clock component (S3 Fig).  Total protein samples were extracted from a continuous culture of ROC75-HA strain exposed to LL conditions. Western blot of α-tubulin is shown as a loading control. The lysate of the WT strain (W) was used as a negative control. (D) Bioluminescence rhythm assay of the ROC75-LUC protein fusion reporter. The traces represent means ± SD of more than five biological replicates. (E) Immunocytochemical staining of ROC75-HA. The cells were harvested from batch TAP cultures of the ROC75-HA strain at L4 To determine the intracellular localization of ROC75, we performed immunocytochemical staining of ROC75-HA. The ROC75-HA signal was detected just inside of the nuclear envelope, which was counterstained with an antibody for the nuclear pore complex (NPC) (Fig 2E). In addition, we detected similar nuclear signals from the ROC75-YFP fusion protein expressed using the same strategy that was employed for ROC75-HA (Fig 2F, S2 Fig). Taken together, these results suggest that ROC75 is a day/subjective day-phase-expressed nuclear-localized protein.

ROC75 associates with the rhythmic circadian clock genes
Because ROC75 has a GARP DNA-binding motif [25], we investigated its association with DNA using chromatin immunoprecipitation followed by deep sequencing (ChIP-seq). We could detect 81 peaks of ROC75-HA associated DNA fragments on the nuclear genome (S1 Table). Most of the peaks were located in the upstream region (36/81) and in the 5 0 -untranslated region (UTR) (31/81) of the target genes (S1 Table, [12]). The peaks were detected on the clock genes (ROC15, ROC40, ROC66, and ROC75 itself) (Fig 3) that are known to show circadian rhythms in their mRNA levels [25]. On the other hand, peaks were not detected on non-rhythmic ROC genes (ROC55 and ROC114 [25]) (Fig 3). The ChIP-seq peak on the ROC40 gene was confirmed by ChIP followed by quantitative PCR (ChIP-qPCR) (S5 Fig). A peak was also detected on ROC59 (S1 Table, Peak No. 5), which is another ROC gene that was identified as the gene responsible for a mutant that showed "long-period" and "delayed-phase" circadian phenotypes [25]. In addition, a peak was detected on CrPRR1, a PRR-like gene encoding a protein similar to the A. thaliana TOC1 [37], but has not yet been described as a clock component in C. reinhardtii (S1 Table, Peak No. 23).
The number of peaks detected was relatively low, probably because of technical reasons. We consider that our ChIP-seq assay covered only highly specific-but not the entire-fraction of ROC75 targets. Therefore, in this study, we focused only on the clock and putative clock genes that were detected.
We tested the DNA-binding activity of the GARP motif of ROC75 by electrophoresis mobility shift assay (EMSA). We used a GST-tagged GARP DNA-binding domain protein of ROC75 (GST-GARP) purified from Escherichia coli (S6A Fig) aGATTTTCt). This binding was weakened by a mutation in the first GAT (S6B Fig,  aCCCTTTTCt) but not by a mutation in its neighboring nucleotides (S6B Fig, cGATCTTCt), and was abrogated by a mutation in both of them (S6B Fig, cCCCCTTCt). This result indicates that the GARP domain of ROC75 can bind to DNA directly in a sequence-specific manner.

ROC75 represses the clock-related genes
We tested the effect of roc75 mutation on the expression of the target genes identified by ChIP-seq analysis at the mRNA level. As described previously [25], the mRNA levels of (4 h after switching-on the lights). Counterstaining with the NPC antibody and merged images are indicated. (F) YFP fluorescence of ROC75-YFP. Batch TAP cultures of the ROC75-YFP strain were transferred to DD conditions, and YFP fluorescence was observed at 18 h after the transfer (subjective midday). White and black bars above the graphs represent the light and dark conditions, respectively. The shaded areas in the main body of the graphs represent the dark period under the LD condition or the time of day corresponding to the dark period for synchronization before the exposure to LL and DD conditions. Representative results are shown (A-F). The reproducibility was confirmed by performing at least two independent experiments (A-F) and by using more than two independent transgenic lines (C-F).
https://doi.org/10.1371/journal.pgen.1008814.g002 ROC15, ROC40, and ROC66 showed robust circadian rhythms, peaking at the subjective night under LL conditions in the WT strain (Fig 4A-4C, LL). Essentially the same rhythmic expression patterns were observed under DD and LD conditions (Fig 4A-4C, DD and LD). In the roc75 mutant, the downregulation of the expression of these genes during the day/subjective day was weak or was almost abolished under all the conditions tested (Fig 4A-4C). These results suggest that ROC75 acts as a repressor of these genes during the day/subjective day. In the case of ROC66, however, the mRNA level was notably low in the mutant during the night/ subjective night, especially under LL and LD conditions (Fig 4C), suggesting the lack of an activation mechanism or suggesting the involvement of additional repressor(s) other than ROC75.
The expression of ROC59 was highly dependent on the light conditions. The expression under DD conditions was one order of magnitude lower than that under the LL conditions, and no rhythmicity was detected both in the WT strain and the roc75 mutant (Fig 4D, DD). Under the LD condition, a peak was observed in the day, and the peak expression level was high in the roc75 mutant (Fig 4D, LD), suggesting that ROC75 acts as a repressor of the lightdependent expression of ROC59 during the day. Under LL conditions, the expression level was high in both the WT strain and roc75 mutant, and the downregulation during subjective day was abolished in the roc75 mutant ( Fig 4D, LL). These results suggest that ROC75 acts as a repressor of the light-dependent expression of ROC59.
The expression of CrPRR1 in the WT strain showed a circadian rhythm, peaking in the evening, under the LD condition and in the early subjective night under LL and DD conditions ( Fig 4E). In the roc75 mutant, the expression during the day/subjective day under LD and LL conditions was higher than in the WT strain (

Acute downregulation of ROC15 and ROC40 mRNAs by light is independent of ROC75
In addition to endogenous oscillations, the ROC15 and ROC40 mRNAs show an acute decline after exposure to light [35,38,39]. To investigate this light response in roc75 to light, cells in the late night phase were exposed to light for 1 h. An acute decline in the expression of these mRNAs was observed in the roc75 mutant as well as in the WT strain ( Fig 5A). The acute decline also occurred after the dawn under the LD condition but the levels of these mRNAs were recovered in the roc75 mutant ( Fig 5B). These results indicate that ROC75 is not essential for the acute downregulation of ROC15 and ROC40 mRNAs by light, but has a role in preventing their reaccumulation during the day.

Rhythms of clock proteins are abolished by roc75 mutation under DD but not under LL and LD conditions
To assess the effect of the roc75 mutation on the rhythms of clock proteins, we measured the bioluminescence rhythms of luciferase reporters translationally fused to ROC15, ROC40, and ROC66 proteins (ROC15-LUC, ROC40-LUC, and ROC66-LUC respectively [35]). Under the LD condition, in sharp contrast to the mRNA data, the bioluminescence rhythms of these reporters were similar to that of WT in the roc75 mutant ( Fig 6A). Under LL conditions, although the rhythm patterns were disturbed in the roc75 mutant, these reporter signals were still rhythmic ( Fig 6B). Under DD conditions, however, these reporter signals were completely arrhythmic in the roc75 mutant, and the bioluminescence levels of ROC15-LUC and ROC40-LUC were constantly high ( Fig 6C). These results suggest that the deregulation of the rhythms of clock mRNAs in the roc75 mutant were reflected in the rhythms of their respective proteins under DD but not under LL and LD conditions. This result is in good agreement with the conditional arrhythmicity of the bioluminescence rhythm in chloroplast reporter strain (Fig 1). RT-qPCR analysis of the mRNA rhythm of the clock-related genes. The cells were harvested from batch HS cultures exposed to LL, DD, and LD conditions. White and black bars above the graphs represent the light and dark conditions, respectively. The shaded areas in the main body of the graphs represent the dark period under the LD condition or the time of day corresponding to the dark period for synchronization before the exposure to LL and DD conditions. The magnified trace of ROC59 expression in WT under the LD condition is shown in blue, and its y-axis is shown on the right (D). Representative results are shown. The reproducibility was confirmed by performing at least two independent experiments. https://doi.org/10.1371/journal.pgen.1008814.g004 Essentially same results were obtained in an in vitro luciferase assay using crude extracts of the reporter cells under DD conditions (S7 Fig), ruling out the possibilities that the results of the in vivo bioluminescence assay were because of the changes in the availability of intracellular ATP, luciferin, magnesium ion, and/or oxygen in this mutant.

Loss of circadian gating of ROC15 reaccumulation in the roc75 mutant
It is obvious that the light-induced acute degradation of ROC15 at dawn occurred normally in the roc75 mutant (Fig 6A, ROC15-LUC), indicating that ROC75 is not involved in this process. Previously, we demonstrated that the recovery of the ROC15 level after degradation was restricted to the subjective night [35]. Interestingly, the recovery was observed throughout the circadian cycle in the roc75 mutant (Fig 7A). In addition, the roc75 mutant failed to keep low ROC15-LUC bioluminescence level during the subjective day when the cells were exposed to a premature darkness (Fig 7B). These results suggest that ROC75 contributes to the circadian gating of ROC15 reaccumulation.

Repression-but not activation-domain fusion is compatible with ROC75
To investigate the molecular function of ROC75, we tested the compatibility of ROC75 with repression-and activation-domain fusion. Fusion to a domain, having the same RT-qPCR analysis of the acute light response. LDsynchronized batch HS cultures were exposed to light (30 μmol/m 2 /s) at the late night (ZT22, 2 h before dawn). One hour after the switching-on of lights, the cells were harvested along with the control cells that were not exposed to light. Data indicate means ± SD of three biological replicates. The asterisks indicate statistically significant differences (Student t-test; P < 0.001). (B) RT-qPCR analysis of the light response at dawn. LD-synchronized batch HS cultures were harvested at 0 (dark), 1, 2, and 4 h after light onset. Data shown are the relative values with respect to dark controls that were kept under dark conditions. White and black bars above the graphs represent the light and dark conditions, respectively. Representative results are shown. The reproducibility was confirmed by performing at least two independent experiments. https://doi.org/10.1371/journal.pgen.1008814.g005 transcriptional function (repression or activation) as ROC75, is expected to enhance or not interfere with its function. Conversely, fusion to a domain having an opposite function is expected to interfere with it. We employed the strong repression domain SRDX [40] and the strong activation domain VP64 [41]. Codon-adapted sequences of the domains for the C. reinhardtii nuclear genome (S8 Fig) were inserted at the end of the coding sequence of the ROC75 genomic DNA fragment to express the C-terminally-tagged version of ROC75 (ROC75-SRDX and ROC75-VP64) (Fig 8A). ROC75-SRDX complemented the arrhythmicity of roc75 mutant at a rate comparable to that of ROC75-HA and ROC75-YFP (Fig 8B, S2 Fig). On the contrary, no transformant with ROC75-VP64 showed restoration of circadian rhythmicity even in transformants expressing the fusion protein at a similar level to ROC75-SRDX (Fig 8B and  8C). Furthermore, upregulation of ROC15 and ROC40 mRNAs was observed in ROC75-VP64 transformants in the day phase (Fig 8D), suggesting that the VP64 domain fusion is functional in C. reinhardtii. These results suggest that ROC75 acts as a transcriptional repressor.

Restoration of ROC75 function resumes the circadian oscillation from subjective dawn
We examined the effect of timing of restoration of ROC75 function in the arrhythmic roc75 mutant in DD. For this purpose, we took advantage of the glucocorticoid receptor (GR) system to control the activity of ROC75. The GR system has been developed to regulate the activity of transcription factors in A. thaliana. Transcription factors fused to the 287-aminio-acid hormone binding domain of GR cannot regulate transcription in the absence of the steroid ligand because of association with a heat-shock protein 90 complex, but the transcriptional activity is restored by adding the ligand dexamethasone (DEX), which induces a conformational change in GR, subsequently releasing the fusion protein from the complex [42].
A codon-adapted sequence of the hormone binding domain of rat GR for the C. reinhardtii nuclear genome (S9 Fig) was inserted at the end of the coding sequence of the ROC75 genomic DNA fragment to express the C-terminally tagged version of ROC75 (ROC75-GR) (Fig 9A). ROC75-GR complemented the roc75 mutant phenotype in the presence of DEX but not in the absence of DEX (Fig 9B). The restoration of the amplitude of bioluminescence rhythm was dependent on the dose of DEX (Fig 9C). These results indicate that the GR system can regulate the activity of ROC75 in C. reinhardtii.
By adding DEX to arrhythmic ROC75-GR cultures under DD conditions, the bioluminescence rhythms were restored (Fig 9D). We added DEX at various times under the DD conditions, and analyzed the circadian phase of the recovered bioluminescence rhythm (Fig 9E, S10  Fig). The results were different between the first day and the second or later days under DD conditions. When DEX was added on the first day, the restored rhythms showed phases relatively close to those of the WT rhythms (Fig 9E and 9F). In other words, the phase was mostly dependent upon the LD cycle before transfer to DD conditions, indicating that the previous light/dark cue is dominant to the ROC75 restoration during this period. In contrast, when DEX was added on the second or later days, the phase of the restored rhythms was dependent upon the timing of DEX addition (Fig 9E and 9F). The peak phase relative to DEX addition was 8-10 h (Fig 9F, left), which was comparable to that relative to the light-on in the WT strain (Fig 9F, right), suggesting that DEX acts as a morning cue in this system.

Roles of ROC75 in the C. reinhardtii circadian clock
In the roc75 mutant under DD conditions, the expression of ROC15 and ROC40 was almost constantly high (Fig 4A and 4B, DD; Fig 6C), similar to their expression state just before the subjective dawn in the WT cells. This suggests that the circadian clock in the roc75 mutant is arrested before the subjective dawn because of the loss of downregulation of these genes by ROC75. This hypothesis is consistent with the result of DEX acting as a morning cue (Fig 9E  and 9F, after the second day under DD conditions). These results suggest that one of the roles of ROC75 is its involvement in the progression of the circadian clock from the subjective night to the subjective day under DD conditions. In addition, the loss of circadian gating of ROC15 reaccumulation (Fig 7) suggests that another role of ROC75 is the maintenance of the "subjective day state" of the circadian clock by preventing reaccumulation of the nighttime clock protein ROC15. This gating might be supported, at least in part, by the repression at the mRNA level by ROC75 (Fig 4A, DD).
In contrast to DD conditions, the circadian rhythms of clock proteins were robust in the LD condition in the roc75 mutant (Fig 6A). It is possible that the acute downregulations in ROC15 and ROC40 mRNAs and ROC15 protein by light (Figs 5 and 7), which are independent of ROC75, generate the robust oscillation of the clock proteins, and compensate the ROC75 functions. Since the mRNA oscillations in the LD condition were weak or not detectable in the roc75 mutant (Fig 4A and 4B, LD) probably due to their rapid recovery after transition to the day phase (Fig 5B), the contribution of the mRNA rhythms on the protein rhythms would be limited. Translational/posttranslational regulations, such as light-dependent degradation of ROC15, probably contribute more to the robust protein oscillations (Fig 6A). However, it is obvious that ROC75 has a role to keep ROC15 and ROC40 mRNA levels low during the day phase after their light-dependent downregulations at dawn (Figs 4A and 4B, 5B). This role is likely to contribute to prevent ROC15 protein accumulation in premature dark conditions (Fig 7B), suggesting that the second role of ROC75, i.e. maintenance of the subjective day state, is significant for the circadian clock even in the LD condition. In natural LD cycles, the light period is often disturbed by darkness due to cloudiness or shade. ROC75 may contribute to keeping time by preventing unexpected ROC15 accumulation in such situations.

Possible additional components of the transcriptional loop of C. reinhardtii circadian clock
ROC59 and CrPRR1 are noteworthy among the target genes of ROC75. Although ROC59 has already been identified as a clock-related gene [25], genetic interaction with ROC75 strengthens the importance of ROC59 in the C. reinhardtii clock. ROC59 encodes a protein harboring several WD40 and tetratricopeptide repeats similar to the evolutionarily conserved antiobesity genes, ASG2 (Altered Seed Germination 2) in higher plants, ADP (antiobesity factor ADI-POSE) in flies, and WDTC1 (WD40 and TPR 1) in humans [25,43]. These genes are known to be involved in fat metabolism [43], but none of them has yet been described as a clock component. Because WDTC1 is known to regulate the transcriptional activities of the genes involved in adipogenesis through histone modifications [44,45], ROC59 might have a function in the epigenetic regulation of the circadian clock in C. reinhardtii.
The CrPRR1 gene encodes a homolog of the O. tauri and A. thaliana clock protein, TOC1 [6,22]. The expression of CrPRR1 peaked at dusk under LD condition and at early subjective night under LL and DD conditions (Fig 4E), indicating that the evening-phased expression is conserved in C. reinhardtii. Investigating whether CrPRR1 is involved in the circadian oscillator function or one of the output pathways from the oscillator in the C. reinhardtii clock would be necessary in understanding the evolutionary history of circadian clocks in green plants.

GARP transcription factors in the plant circadian clocks
In this study, we demonstrate that ROC75 is a daytime component of the C. reinhardtii clock that represses the mRNA expression of night-phased clock genes (e.g., ROC15 and ROC40). This is in contrast to the case of A. thaliana LUX (PCL1) which is a nighttime repressor [12]. The features of ROC75 and LUX (PCL1) are summarized in Table 1. They also show sharp contrasts in their mutant phenotypes: An arrhythmic phenotype was observed in the roc75 mutant under DD conditions (Fig 1B and 1D) [25] but in lux/pcl1 mutants, it is observed under LL conditions [10,11]. In addition, a weak rhythm phenotype was observed under LL conditions in roc75 (Fig 1A and 1D) but under DD conditions in the CCR2::LUC reporter in lux mutant [10]. On the other hand, both the mutants (roc75 and lux) showed normal rhythm under LD conditions at least in some of the processes (Figs 1C, 1D and 6A) [10]. Meanwhile, O. tauri genome [46] has GARP protein genes, and one of them encodes a LUX-like protein (ostta12g01420). Although mutant phenotypes and the effect of misexpression have not been reported to date, the mRNA expression of this gene is evening-phased under LD conditions [47], similar to that of LUX (PCL1) and BOA (NOX) [10,11,15], in contrast to that of ROC75 (Fig 2A, 2C and 2D) [25]. One possibility is that C. reinhardtii has developed a contrasting mechanism of operating the circadian clock to that of A. thaliana and O. tauri. Consistent with this notion, the effect of changes in the casein kinase 1 activity on the circadian clock also shows a sharp contrast; the circadian rhythm is shortened by an RNAi-mediated knockdown in C. reinhardtii [48], whereas it is lengthened by specific inhibitor treatments or misexpression in A. thaliana and O. tauri [49,50].
It is still unclear to what extent the C. reinhardtii clock is similar to the other plant clocks in the network architectures of the circadian transcriptional loops. This study indicates that at least the relationship between GARP proteins (ROC75 and LUX [PCL1]) and MYB transcription factor genes (ROC40 and CCA1/LHY) is different between C. reinhardtii and A. thaliana. In C. reinhardtii, ROC40 is a direct target of ROC75, and therefore, its expression is constantly high in the roc75 mutant (Fig 4B). In contrast in A. thaliana, the expression of CCA1/LHY is indirectly regulated by LUX (PCL1) via PRR9 [12]. Since PRR9 is a transcriptional repressor [14], LUX (PCL1) act as an indirect activator of CCA1/LHY. The expression of CCA1/LHY is, therefore, constantly low in lux/pcl1 mutants [10,11].
Since Prasinophyceae is an early-diverging class within the green plant lineage [16], the reduced circadian system in O. tauri [20,21] seems to have retained the primitive features of plant circadian clocks. The evening-loop may be the original position of GARP transcription factor in the plant circadian clocks, and the day-loop, such as ROC75, may be a newly acquired loop in the Chlorophyceae, including C. reinhardtii. Further comparative analyses among the three models, C. reinhardtii, O. tauri and A. thaliana, will provide insight into the evolution of the circadian clocks in the green plant lineage.
The ROC75-HA, ROC75-YFP, ROC75-SRDX, ROC75-VP64 and ROC75-GR strains were generated by transformation of the roc75 mutant. To generate transforming DNAs, artificially synthesized DNA fragments encoding HA, YFP, SRDX, VP64, and GR were cloned into the FseI site of the pSI103PL/ROC75 plasmid [25] via ligation or In-Fusion assembly. The plasmids were digested with PacI, and the fragments containing the tagged ROC75 gene and the aphVIII selectable marker gene were used for transformation. The transformation was performed as described previously [35,52,53].

Bioluminescence rhythm assay
The bioluminescence rhythm of spot cultures of reporter cells in white 96-well plates was monitored as described previously [25] except for the experiment in Fig 8B where we used spot cultures in black 24-well plates and a bioluminescence monitoring apparatus CL24-LIC (Churitsu Electric Corp., Nagoya, Japan). The circadian parameters (period, phase, and amplitude) were calculated by the cosinor method of the Rhythm Analyzing Program [54]. The amplitude was estimated from the overall average best-fitting cosinor curves between the second and the fifth day of bioluminescence monitoring. To monitor the clock protein-luciferase fusion reporters (Figs 2D, 6 and 7 and S3 Fig), we used AkaLumine-HCl (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) [55] as the substrate because the bioluminescence rhythm was more stable against environmental disturbances (e.g., light on/off, drying of agar, vibration of culture plates).

Genetic crosses
Genetic crosses were carried out as described previously [25]. The progeny were screened by bioluminescence and antibiotic resistance, and the genotype was confirmed by genomic PCR.
2) Batch HS culture: Two flasks of cells that were 12-h out-of-phase were prepared. The light and temperature schedules for LL, DD, and LD conditions are summarized in S11A-S11C Fig. The cells were inoculated into high-salt (HS) medium [56] at a concentration of 2 × 10 5 cells/mL and incubated at 24˚C under LL conditions (30 μmol/m 2 /s) for 3 days. After synchronization of the circadian clock employing both a temperature shift to 17˚C and a 12-h dark exposure, the cultures were kept under the LL, DD, or LD condition. The cells were harvested at the indicated time points (24 to 72 h). The overlapping time points between the two flasks (36,48, and 60 h) were used to validate the flask-to-flask variation. The sample collection during the dark period was done under safe, very low-intensity (less than 0.01 μmol/m 2 /s) blue LED light. Air was bubbled through the cultures exposed to the DD condition to ensure a stable circadian rhythm. No growth was observed even in the LL cultures (S11D Fig), ruling out the possibility of cell cycle-dependent rhythms. 3) Continuous culture: The continuous culture was carried out as described previously [25].

Western blot
Continuous culture (5 mL, 1-2x10 7 cells) was harvested by centrifugation. Protein extraction and Western blot were carried out as described previously [39].

RT-qPCR
Batch HS culture (5 mL, 1-2x10 7 cells) was harvested by centrifugation. RNA extraction and RT-qPCR was carried out as described previously [39]. Data were normalized with respect to the expression level of RCK1. The sequences of primers used in the study are provided in S2 Table. YFP fluorescence imaging A small aliquot (few microliters) of the batch TAP culture was dropped onto a cover slip and the drop was covered with a small piece (approximately 10 mm × 10 mm) of plastic wrap. C. reinhardtii cells were concentrated and immobilized in the wrinkles formed in the plastic wrap. The YFP fluorescence of the immobilized cells was observed with an inverted fluorescence microscope (IX70, Olympus) equipped with a CCD camera (DS-5M, Nikon). The fragment libraries were constructed using the SOLiD fragment library construction kit (Thermo Fisher Scientific) according to the manufacture's protocol, except for the purification steps that were carried out using the Agencourt AMPure XP Kit (Beckman Coulter). The pooled library was sequenced by multiplex paired-end sequencing on a flow cell using the 5500xl SOLiD system (Life Technologies, Waltham, MA, USA). The sequence reads were mapped to the reference genome of C. reinhardtii v 5.5 (Joint Genome Institute) [57] using Bioscope (version 1.2, Applied Biosystems SOLiD System). The total number of reads and mapped reads in each experiment are summarized in S3 Table. Mapping data has been registered in the DDBJ Sequence Read Archive (https://www.ddbj.nig.ac.jp/dra/index.html) under the accession numbers from DRR203090 to DRR203093. Peaks were detected by the peak calling algorithm for ChIP-seq analysis of CLC Genomics Workbench (version 7.5.1, Qiagen). To confirm the specificity of the detected peaks, we selected the peaks that were common in two biological replicates, and all the selected peaks were visually confirmed by comparing with the input DNA controls.

ChIP-seq and ChIP-qPCR
Quantitative PCR of ChIP DNA was performed using Fast SYBR Green I Master Mix (Thermo Fisher Scientific) and StepOnePlus (Thermo Fisher Scientific) according to the manufacture's procedure. The primers used are provided in S2 Table.

Preparation of the Glutathione-S-transferase (GST)-GARP protein and EMSA
The GST-GARP protein was expressed in the E. coli strain BL21 (DE3) cultured in LB and Terrific Broth media, supplemented with ampicillin (50 μg/mL). The cells were harvested by centrifugation, resuspended in 50 mM Tris-HCl buffer (pH 8.0) containing 50 mM NaCl, 1 mM EDTA, and 1 mM dithiothreitol (DTT), and disrupted by sonication. The lysates were clarified by centrifugation, and a GST-fused form of GARP was prepared from the lysates using Glutathione Sepharose 4B (GE Healthcare) and dialyzed for 12 h against 1 L of 20 mM Tris-HCl buffer (pH 8.0) containing 10 mM NaCl, 0.5 mM EDTA, and 1 mM DTT. The dialysates were applied to a Hitrap DEAE FF column (GE Healthcare) to purify the GST-GARP, and the proteins were further purified by ion-exchange chromatography on a MonoQ HR 5/5 column (GE Healthcare). The GST-GARP was then applied to a Hiload Superdex 75 10/300 GL column (GE Healthcare), equilibrated with 20 mM Tris-HCl (pH 7.5) containing 150 mM NaCl and 1 mM DTT. The concentration of proteins was determined using the Bio-Rad Protein Assay (Bio-Rad Laboratories); bovine serum albumin was used as a standard.

In vitro luciferase assay
The cells were prepared in the same way as described for the bioluminescence rhythm assay. The cell spot on agar piece was flash frozen in liquid nitrogen in a 1.5-mL tube and stored at -80˚C. The cell spots were lysed by adding 150 μL of Steady-Glo Luciferase Assay Reagent (Promega) and vortexed for 10 min. After the agar piece settled down, 80 μL of the supernatant was used for monitoring the luminescence in a multilabel plate reader (ARVO X4, PerkinElmer).